experiments
/
STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove `_enable_legacy_iterators` flag in TPUStrategy. 

PiperOrigin-RevId: 310656609
Change-Id: I0d4eaa8139689df1b6f644c87b1cd4028831dd52
This commit is contained in:
Ruoxin Sang 2020-05-08 17:09:34 -07:00 committed by TensorFlower Gardener
parent ebd34b3dc1
commit 4204c5f856
1 changed files with 0 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/python/distribute/tpu_strategy.py
View File

@ -140,10 +140,6 @@ class TPUStrategy(distribute_lib.Strategy):
distribute_lib.distribution_strategy_replica_gauge.get_cell(
"num_replicas_per_worker").set(self.extended.num_replicas_per_host)
# TODO(b/155193424): Enable OwnedMultiDeviceIterator on TPU Pod.
if self.extended.num_hosts > 1:
self._enable_legacy_iterators = True
# TODO(cjfj): Modify `_call_for_each_replica` in `TPUExtended` such that this
# can use the default implementation.
# This implementation runs a single step. It does not use infeed or outfeed.